summaryrefslogtreecommitdiff
path: root/caffe2/utils
AgeCommit message (Expand)AuthorFilesLines
2017-08-01Fix #997Yangqing Jia1-0/+7
2017-07-26Move cpuid ctor to .ccYangqing Jia2-71/+80
2017-07-26MKL code moveYangqing Jia9-2194/+0
2017-07-26CMake updates.Yangqing Jia1-19/+0
2017-07-26MKLDNN bugfixYangqing Jia1-11/+11
2017-07-25CosineSimilarity GPUWojciech Glogowski3-0/+57
2017-07-23Add proper cpuid support.Yangqing Jia3-0/+222
2017-07-21comment out unused parametersVictor Gao7-28/+43
2017-07-21added sincos function to caffe2/utils/mathMatt Uyttendaele3-0/+45
2017-07-19Nuke arg_helper() in OperatorBaseJunjie Bai1-10/+38
2017-07-18Sync mobile codebase changes back to fbcodeJon Morton3-66/+122
2017-07-17Rename def() to debug_def()Junjie Bai1-1/+2
2017-07-17Use shared_ptr to share OperatorDef across threadsJunjie Bai2-13/+12
2017-07-17remove unused parameters in math_cpu.ccVictor Gao1-149/+147
2017-07-17Change Allocator interface to return deleterDmytro Dzhulgakov1-4/+1
2017-07-07Revert threadpool minWorkSize change on iOSAndrew Tulloch1-0/+4
2017-07-07One more proto_utils.h fixDmytro Dzhulgakov1-1/+1
2017-07-06Reduce default ThreadPool min work size (~25% speedup for segmentation on S7).Andrew Tulloch1-1/+1
2017-07-05Small improvements to CreateOperatorDefDmytro Dzhulgakov1-19/+28
2017-07-05Use the same schema of switching to device reduce sum for SumSqrElementsJunjie Bai3-68/+120
2017-07-03Cos, Sin, and Abs operatorsMarat Dukhan3-0/+18
2017-06-30Use cub::DeviceReduce for faster math::Sum CUDA versionJunjie Bai3-9/+95
2017-06-21Remove unused thrust headers from math_gpu.Ahmed Taei1-5/+0
2017-06-21Revert D5273337: [caffe2] Pare down on excessive futex() syscalls from the DA...Aapo Kyrola1-11/+0
2017-06-19Pare down on excessive futex() syscalls from the DAGNet executorJames Reed1-0/+11
2017-06-17Implement TopKOp for GPUJeff Johnson4-2/+405
2017-06-16Disable smart_tensor_printer_test on OSXLuke Yeager1-0/+4
2017-06-14MaxReduction ops GPU implementation.Ahmed Taei1-0/+77
2017-06-08Fixed thread safety issues in ImageInputOpRomain Cledat1-6/+33
2017-05-31Disable stacktrace on fatal signal by defaultDan Zimmerman3-52/+118
2017-05-26Use unwind functions instead of backtrace to attempt to be more portableDan Zimmerman1-6/+18
2017-05-25Add asan no sig tests, move fatal sig tests thereDan Zimmerman1-9/+14
2017-05-24Clean up binary build cmake scriptYangqing Jia2-39/+30
2017-05-22Add stack traces on fatal signalsDan Zimmerman4-61/+449
2017-05-17fix bad conversion to float in cpu_half2floatLukasz Wesolowski2-8/+23
2017-04-28Support fp16 output from ImageInputOpSimon Layton2-2/+34
2017-04-27fix curand odd-sized workaroundKevin Matzen1-2/+1
2017-04-25MKL related files with review comments incorporatedintel2-0/+195
2017-04-24guard against apple platformsYangqing Jia1-4/+7
2017-04-24fix thread_local bugYangqing Jia1-1/+4
2017-04-24caffe2: smart_tensor_printerJanusz Kudelka3-0/+168
2017-04-19fp16 support for FullyConnected op(Fixed)Ahmed Taei6-238/+632
2017-04-17Revert D4871248: [caffe2][PR] fp16 support for FullyConnected opAapo Kyrola6-632/+238
2017-04-17fp16 support for FullyConnected opSimon Layton6-238/+632
2017-04-14Add __builtin_cpu_supports function def in windowsDongsheng Fang1-0/+19
2017-04-12Add math::Gemv<double, CUDAContext> by cublas::cublasDgemvYiming Wu2-0/+85
2017-04-10SumSqrElementsAapo Kyrola3-8/+40
2017-04-08DuplicateOpKittipat Virochsiri1-12/+12
2017-04-04Constant string is generated from Protobuf instead of ThriftFei Sun1-1/+1
2017-04-03Add default value for GetRepeatedFieldAlexander Sidorov2-24/+27
generated by cgit v1.2.3 (git 2.25.1) at 2024-11-30 07:29:59 +0000